EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

4353. Uncertain Unit Commitment For The Operation Of Hybrid Power Plants Under Uncertainty

Invited abstract in session TC-22: Stochastic models in energy systems planning and operations, stream Energy Management.

Tuesday, 12:30-14:00
Room: 81 (building: 116)

Authors (first author is the speaker)

1. Nouha Dkhili
TotalEnergies
2. vincent leclere
CERMICS, Ecole des Ponts
3. Marco Marchese
OT/RD/H&S/CSO, TotalEnergies
4. Anna Robert
TotalEnergies R&D
5. Vitor Luiz Pinto de Pina Ferreira
CERMICS, École des Ponts ParisTech

Abstract

The increased deployment of renewable energies introduces intermittency in power generation that was reliably dispatched in the past. The conventional Unit Commitment problem in energy system management consists in optimizing the operation schedule of thermal units to meet the load at minimum cost and CO2 emissions. However, Uncertain Unit Commitment (UUC) models extend this framework by integrating uncertainties stemming from generation, load fluctuations, pricing, etc. These extensions require decomposition-based computational methods to solve these problems efficiently and accurately. TotalEnergies collaborates with Ecole des Ponts (CERMICS lab) to focus on an isolated industrial energy-intensive site, initially with a known load. The energy demand must be met by internal resources: thermal generators, PV panels, and batteries. The complexity of this problem stems from the interplay of factors like storage and ramping constraints, interstage coupling, binary variables, and the intermittency of solar energy. Ensuring continuous load fulfillment remains imperative for safety and reliability. Yet, an overly cautious approach can result in high production costs and CO2 emissions. We study the risk-neutral and risk-averse approaches via the stochastic and robust settings, respectively. By separating the information structures behind binary and continuous variables, we can decouple them in the optimization problem and use cut-generation algorithms to solve both problems.

Keywords

Status: accepted

Back to the list of papers